adds scripts

scripts/01_download_data.sh

@@ -0,0 +1,62 @@
+
+mkdir -p data/resources
+
+cd data/resources
+
+# panel of normals
+
+aws s3 \
+--no-sign-request --region eu-west-1 \
+cp \
+s3://ngi-igenomes/igenomes/Homo_sapiens/GATK/GRCh38/Annotation/GATKBundle/1000g_pon.hg38.vcf.gz \
+.
+
+aws s3 \
+--no-sign-request --region eu-west-1 \
+cp \
+s3://ngi-igenomes/igenomes/Homo_sapiens/GATK/GRCh38/Annotation/GATKBundle/1000g_pon.hg38.vcf.gz.tbi \
+.
+
+cd ../
+mkdir reference
+
+cd reference
+
+# exome intervals
+wget https://genomedata.org/pmbio-workshop/references/exome/chr6_and_chr17/exome_regions.bed
+wget https://genomedata.org/pmbio-workshop/references/exome/chr6_and_chr17/exome_regions.bed.interval_list
+
+# reference genome
+wget https://genomedata.org/pmbio-workshop/references/genome/chr6_and_chr17/ref_genome.tar
+tar xvf ref_genome.tar
+rm ref_genome.tar
+
+# reads
+cd ../
+mkdir reads
+cd reads
+
+wget https://genomedata.org/pmbio-workshop/fastqs/chr6_and_chr17/Exome_Norm.tar
+wget https://genomedata.org/pmbio-workshop/fastqs/chr6_and_chr17/Exome_Tumor.tar
+
+tar xvf Exome_Norm.tar
+rm Exome_Norm.tar
+tar xvf Exome_Tumor.tar
+rm Exome_Tumor.tar
+
+mv Exome_Norm/Exome_Norm_R1.fastq.gz normal_R1.fastq.gz
+mv Exome_Norm/Exome_Norm_R2.fastq.gz normal_R2.fastq.gz
+
+mv Exome_Tumor/Exome_Tumor_R1.fastq.gz tumor_R1.fastq.gz
+mv Exome_Tumor/Exome_Tumor_R2.fastq.gz tumor_R2.fastq.gz
+
+rm -r Exome_Norm
+rm -r Exome_Tumor
+
+# subset vcf
+cd ../resources
+bcftools view -Oz -r chr6,chr17 af-only-gnomad.hg38.vcf.gz > af-only-gnomad.hg38.subset.vcf.gz
+bcftools index --tbi af-only-gnomad.hg38.subset.vcf.gz
+
+bcftools view -Oz -r chr6,chr17 1000g_pon.hg38.vcf.gz > 1000g_pon.hg38.subset.vcf.gz
+bcftools index --tbi 1000g_pon.hg38.subset.vcf.gz

scripts/02_build_reference.sh

@@ -0,0 +1,6 @@
+
+REFDIR=/config/data/reference/
+
+mkdir -p "$ALIGNDIR"
+
+bwa index "$REFDIR"/ref_genome.fa

scripts/03_alignment.sh

@@ -0,0 +1,18 @@
+
+REFDIR=/config/data/reference/
+READDIR=/config/data/reads
+ALIGNDIR=/config/data/alignments
+
+mkdir -p "$ALIGNDIR"
+
+for sample in tumor normal
+do
+    bwa mem \
+    "$REFDIR"/ref_genome.fa \
+    "$READDIR"/"$sample"_R1.fastq.gz \
+    "$READDIR"/"$sample"_R2.fastq.gz \
+    2> "$ALIGNDIR"/$sample.bwa.log \
+    | samtools sort \
+    | samtools view -bh \
+    > "$ALIGNDIR"/"$sample".bam
+done

scripts/04_add_readgroups.sh

@@ -0,0 +1,15 @@
+#!/usr/bin/env bash
+
+ALIGNDIR=/config/data/alignments
+
+for sample in tumor normal
+do
+    gatk AddOrReplaceReadGroups \
+    --INPUT "$ALIGNDIR"/"$sample".bam \
+    --OUTPUT "$ALIGNDIR"/"$sample".rg.bam \
+    --RGLB "$sample" \
+    --RGPU HWI-ST466.C1TD1ACXX \
+    --RGPL ILLUMINA \
+    --RGSM "$sample" \
+    --RGID HWI-ST466.C1TD1ACXX."$sample"
+done

scripts/05_mark_duplicates.sh

@@ -0,0 +1,13 @@
+#!/usr/bin/env bash
+
+ALIGNDIR=/config/data/alignments
+
+for sample in tumor normal
+do
+    gatk MarkDuplicates \
+    --INPUT "$ALIGNDIR"/"$sample".rg.bam \
+    --OUTPUT "$ALIGNDIR"/"$sample".rg.md.bam \
+    --METRICS_FILE "$ALIGNDIR"/marked_dup_metrics_"$sample".txt 
+
+    samtools index "$ALIGNDIR"/"$sample".rg.md.bam
+done

scripts/06_run_mutect2.sh

@@ -0,0 +1,18 @@
+#!/usr/bin/env bash
+
+ALIGNDIR=/config/data/alignments
+REFDIR=/config/data/reference
+RESOURCEDIR=/config/data/resources
+VARIANTDIR=/config/data/variants
+
+mkdir -p $VARIANTDIR
+
+gatk Mutect2 \
+-R "$REFDIR"/ref_genome.fa \
+--intervals "$REFDIR"/exome_regions.bed.interval_list \
+-I "$ALIGNDIR"/tumor.rg.md.bam \
+-I "$ALIGNDIR"/normal.rg.md.bam \
+-normal normal \
+--germline-resource "$RESOURCEDIR"/af-only-gnomad.hg38.subset.vcf.gz \
+--panel-of-normals "$RESOURCEDIR"/1000g_pon.hg38.subset.vcf.gz \
+-O "$VARIANTDIR"/somatic.vcf.gz

scripts/07_get_pileups.sh

@@ -0,0 +1,14 @@
+#!/usr/bin/env bash
+
+ALIGNDIR=/config/data/alignments
+RESOURCEDIR=/config/data/resources
+VARIANTDIR=/config/data/variants
+
+for sample in tumor normal
+do
+    gatk GetPileupSummaries \
+    -I "$ALIGNDIR"/"$sample".rg.md.bam \
+    -V "$RESOURCEDIR"/af-only-gnomad.hg38.vcf.gz \
+    -L "$RESOURCEDIR"/af-only-gnomad.hg38.vcf.gz \
+    -O "$VARIANTDIR"/"$sample".pileups.table
+done

scripts/08_calculate_contamination.sh

@@ -0,0 +1,10 @@
+
+#!/usr/bin/env bash
+
+ALIGNDIR=/config/data/alignments
+RESOURCEDIR=/config/data/resources
+
+gatk CalculateContamination \
+-I "$VARIANTDIR"/tumor.pileups.table \
+-matched "$VARIANTDIR"/normal.pileups.table \
+-O "$VARIANTDIR"/ontamination.table

scripts/start_container.sh

@@ -0,0 +1,3 @@
+
+cd /Users/geertvangeest/Documents/repositories/cancer-variants-training
+docker run --rm -v $PWD:/config -p 8443:8443 geertvangeest/cancer-variants-vscode:latest